CSLI LinGO Lab

This page provides information about the CSLI Linguistic Grammars Online (LinGO) Lab at Stanford University.

The CSLI LinGO Lab is committed to the development of linguistically precise grammars based on the HPSG framework, and general-purpose tools for use in grammar engineering, profiling, parsing and generation. Early work in the CSLI LinGO Lab focused on the construction of a general-purpose grammar of English in the form of the English Resource Grammar (or ERG), and on further development of the LKB grammar engineering system. The LKB was also used at CSLI as the testbed for a number of teaching grammars and smaller-scale grammars for other languages including Japanese and Spanish.

Since 2003, LinGO has worked closely with the Language Technology Group (LTG) at the University of Oslo, where the ERG has been used for generation in the Norwegian LOGON machine translation project. The collaboration was formalized in 2007 as a research partnership between the two sites, leading to the joint WeScience initiative—applying the ERG and other DELPH-IN technology to scholarly literature in our own field. The partners coordinate their resource and technology development, exchange staff and students, and publish jointly.

Beginning in 2009, the ERG and the PET parser (a DELPH-IN resource) have been included as components of Stanford's Education Program for Gifted Youth (EPGY) software, in their Language Arts and Writing online course. Using an extended grammar rule inventory that admits mildly ungrammatical structures (cf. Bender et al. (2004)), the ERG identifies syntactic errors in sentences composed by elementary school students, enabling EPGY software to suggest improvements to the students as they write. The system is currently in weekly use by some 20,000 students in the United States.

The ERG has played a central role in research on multiword expressions (MWEs) jointly funded by the NSF and NTT Communication Science Laboratories, Nippon Telegraph and Telephone Corporation. It has also been deployed in the construction of a novel, rich and dynamic treebank for HPSG, the LinGO Redwoods initiative. The ERG has also been used for grammar-checking in technical manuals, in an experiment funded by the DFKI Language Technology Lab. Together with colleagues at the University of Washington and others in the DELPH-IN network, we have brought together our grammar writing experience across a variety of languages in devising a Grammar Matrix to aid in the development of broad-coverage, precision, implemented grammars for natural languages.

The CSLI LinGO Lab is one of the founding members of the international DELPH-IN (Deep Linguistic Processing with HPSG Initiative), a network of academic and commercial research institutions that target applications of precise, linguistic NLP. HPSG and Minimal Recursion Semantics (MRS) provide the common scientific framework for DELPH-IN members, and many of the LinGO resources are part of the DELPH-IN open-source repository.

In past projects, the ERG was used in the Verbmobil machine translation system, and in an NSF-funded project on computer-aided speech generation for people who cannot speak because of disability.

Last modified: Fri Dec 18 19:48:42 PST 2009